Compilation failure with simd_rem

[AdamNiederer]

Hello,
stdsimd fails to build on my machine after 663cf64.

Adam, [21.11.17 00:49]
cargo build 
   Compiling stdsimd v0.0.3 (file:///home/adam/Programs/stdsimd)
error[E0441]: unrecognized platform-specific intrinsic function: simd_rem
  —> src/simd_llvm.rs:28:5
   |
28 |     pub fn simd_rem<T>(x: T, y: T) -> T;
   |     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

error: aborting due to previous error

error: Could not compile stdsimd.

I'm on an i3 5005u (+avx2). stdsimd compiles without issue on my 3570k (+avx). I'm using LLVM 5.0.0 on both, and changing the target-cpu has no effect on either machine.

[gnzlbg]

@AdamNiederer which Rust version do you have?

simd_rem requires a nightly that's newer than 16 days, and the AVX-512 intrinsics require a nightly newer than a couple of days.

Do you have any RUSTFLAGS= set or are you just doing cargo build ? Could you maybe show the output of cargo build --verbose ?

[AdamNiederer]

Ah, such is life on the bleeding edge. Updating rustc fixed the issue; thanks.

[gnzlbg]

@AdamNiederer btw I added these intrinsics to rustc and stdsimd because I saw that you were using the vector type "operators" in faster. Since that commit, all useful operators should be implemented and working properly. If you miss any please ping me either here or in the rustc repo. Happy hacking!

---

I'll let you know once we release the next version of stdsimd so that you can migrate to a non-moving target.

[AdamNiederer]

Awesome, thank you! I'm trying to hack together the gather intrinsics here so I can get a a proper stream swizzling API working in faster right now. I should have those ready to merge in a few days.